Observation of an Oxonium Ion Intermediate in Ethanol Dehydration to Ethene on Zeolite

ARTICLE

https://doi.org/10.1038/s41467-019-09956-7 OPEN Observation of an oxonium ion intermediate in ethanol dehydration to ethene on zeolite

Xue Zhou1,2,4, Chao Wang1,4, Yueying Chu1,4, Jun Xu1,3, Qiang Wang 1, Guodong Qi1, Xingling Zhao1,2, Ningdong Feng1 & Feng Deng1

Zeolite-catalyzed dehydration of ethanol offers promising perspectives for the sustainable production of ethene. Complex parallel-consecutive pathways are proposed to be involved in

1234567890():,; the reaction network of ethanol dehydration on zeolites, where the initial step of ethanol dehydration is still unclear particularly for the favorable production of ethene at lower temperature. Here we report the observation of a triethyloxonium ion (TEO) in the dehy- dration of ethanol on zeolite H-ZSM-5 by using ex situ and in situ solid-state NMR spec- troscopy. TEO is identified as a stable surface species on the working catalyst, which shows high reactivity during reaction. Ethylation of the zeolite by TEO occurs at lower temperature, leading to the formation of surface ethoxy species and then ethene. The TEO-ethoxide pathway is found to be energetically preferable for the dehydration of ethanol to ethene in the initial stage, which is also supported by theoretical calculations.

1 National Centre for Magnetic Resonance in Wuhan, State Key Laboratory of Magnetic Resonance and Atomic and Molecular Physics, Key Laboratory of Magnetic Resonance in Biological Systems, Wuhan Institute of Physics and Mathematics, Chinese Academy of Sciences, Wuhan 430071, China. 2 University of Chinese Academy of Sciences, Beijing 100049, China. 3 Wuhan National Laboratory for Optoelectronics, Huazhong University of Science and Technology, Wuhan 430074, China. 4These authors contributed equally: Xue Zhou, Chao Wang, Yueying Chu. Correspondence and requests for materials should be addressed to J.X. (email: [email protected]) or to F.D. (email: [email protected])

NATURE COMMUNICATIONS | (2019) 10:1961 | https://doi.org/10.1038/s41467-019-09956-7 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09956-7

thene is one of the most important commodity chemicals, Diethyl ether (DEE) is concurrently produced with ethene, and Ewhich is currently produced by cracking processes from the kinetic and spectroscopic studies revealed that intermolecular petroleum. Due to the large availability of bioethanol from dehydration of ethanol leads to DEE via dimeric ethanol species renewable biomass sources and the dwindling of fossil resource1, or reaction of an ethoxy group with undissociated ethanol on the conversion of ethanol to ethene and higher hydrocarbons now zeolites13,22,23. The following cracking of DEE produces ethene, is receiving increasing attention from both academia and which occurs preferably at lower reaction temperatures as com- industry2,3. Among of the heterogeneous catalysts studied, zeo- pared to the unimolecular dehydration of ethanol to ethene14,24. lites especially ZSM-5 or modified ZSM-5 are the promising Note that the ethoxy species was generally proposed to be the catalysts as they can be tuned to exhibit higher activity than the direct precursor to ethene. However, the generation of such traditional alumina-based catalysts4–9: the reaction temperature is species is still not well understood. Thus, although different lower and a low concentration of ethanol aqueous solution can be reaction schemes have been proposed, the detailed mechanism of used. The formation of ethene is the first step in ethanol dehy- the initial step of ethanol dehydration remains elusive, particu- dration; subsequent polymerization, cracking and aromatization larly for the favorable production of ethene from DEE at lower by the secondary reaction of ethene leads to the formation of temperature. longer-chain hydrocarbons10, very similar to the formation of Here we report the observation and identification of reaction hydrocarbons in the conversion of methanol over zeolites11,12. intermediates in ethanol dehydration to ethene over zeolite The detailed knowledge of the reaction mechanism of ethene H-ZSM-5. The ex situ and in situ solid-state NMR spectroscopy formation in ethanol dehydration is important not only for allows for exploration of the dehydration of ethanol at the initial understanding ethanol dehydration as a model reaction of alcohol stage. A triethyloxonium ion (TEO) is observed as a potential conversion but also for the development of improved catalysts for active intermediate on the working catalyst. We provide evidence light olefins production. for the involvement of TEO in the formation of surface ethoxy The formation of ethene in the initial dehydration process has species and then ethene on the zeolite. The elementary reactions been experimentally13–15 and theoretically16–19 investigated. A leading to the formation of ethene are identified by combining parallel-consecutive route was proposed3: dehydration of ethanol experiments and theoretical calculations, which suggests that a to ethene and diethyl ether in parallel and a consecutive TEO-mediated reaction route is operative for the dehydration of decomposition of diethyl ether to ethene. In the unimolecular ethanol to ethene. dehydration of ethanol to ethene, it was proposed that the acidic proton transfers from the zeolite to the OH group on ethanol, which subsequently releases water to form an intermediate such Results as carbocation or an alkoxide bound to zeolite. Solid-state NMR Observation of TEO. H-ZSM-5 (Si/Al = 11.5, Zeolyst) was used and infrared spectroscopic studies16,20,21 confirmed the forma- for the dehydration of ethanol. The XRD and 27Al and 29Si solid- tion of surface-bound ethoxy species as a stable intermediate in state NMR spectra show structure information of the H-ZSM-5 ethanol dehydration of on H-ZSM-5 and HY zeolites. Ethene and catalyst (Supplementary Fig. 1 and 2). The acidic property was hydrocarbons are formed by the subsequent decomposition of the examined by FT-IR and NH3-TPD (Supplementary Fig. 3 and 4). surface ethoxy species. The reactions were conducted in a pulse-quench reactor25,in which ethanol was pulse-injected at temperatures ranging from 160 to 350 °C and allowed to react for 4 s under continuous He a carrier gas flow before the reaction was thermally quenched by 160°C pulsing liquid nitrogen onto the catalyst bed. The reaction fl X8 ef uent was analyzed by GC (Supplementary Fig. 5). Ethene is the * only product at 160–200 °C, indicating that the dehydration of ethanol to ethene is favorable at lower temperature. With 69.0 increasing reaction temperature, the secondary reaction of ethene b + 200°C 62.5 fi

85.0 leads to the formation of C3 products including ole ns and X8 * aromatics. Figure 1 shows the ex situ 13C CP/MAS NMR spectra of trapped products from the dehydration of 13C-1-enriched ethanol ° 13 13 – c 250 C (CH3 CH2OH, 99% C) on H-ZSM-5 at 160 350 °C for 4 s. X8 * Ethanol and DEE are observed as reflected by the strong signals at 62.5 and 69.0 ppm respectively, due to the 13C-enriched d 300°C methylene carbons; the methyl carbons produce the weak signals at 12–17 ppm. The adsorbed species formed at 200 °C were X8 * further analyzed by two-dimensional (2D) 1H–13C HETCOR NMR experiment (Supplementary Fig. 6). A weak signal at e 350°C 72.8 ppm that was seriously overlapped by the ethanol signal (69.0 ppm) in the 1D 13C NMR spectra was clearly discerned in 147.3

156.3 the 2D spectrum, which can be ascribed to the methylene carbon 243.6 255.7 249.6 32.2 29.2 47.6 126.5 9.7 X8 17.0 of ethoxy group. The formation of the alkoxy species was well * established in methanol and ethanol dehydration on acidic zeolites. Hunger et al.20 reported the surface ethoxy species from 250 200 150 100 50 0 ppm ethanol dehydration on HY zeolite, which has a chemical shift of 72.6 ppm. Fig. 1 13C NMR analysis of the formed surface species on zeolite. 13C CP/ Interestingly, a well-resolved signal appears at 85.0 ppm (Fig. 1) MAS NMR spectra of trapped species on H-ZSM-5 obtained from reaction and grows up at intermediate temperature. The oligomeric alkoxy 13 of CH3 CH2OH for 4 s at different temperatureof 160 °C–350 °C (a-e). species formed by the oligomerization of ethene on acidic zeolites Asterisks denote spinning sidebands were previously proposed to produce broad signals at 78.0 to

2 NATURE COMMUNICATIONS | (2019) 10:1961 | https://doi.org/10.1038/s41467-019-09956-7 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09956-7 ARTICLE

20,26 90.0 ppm in the literatures . Since there is no any C3+ species TEO and the ethoxy species are solely formed and located on formed at this reaction condition, the oligomerization of ethene Brønsted acid sites for charge compensation, while DEE and did not likely occur in our case. Furthermore, no methylene ethanol are trapped over zeolite via either chemical or physical signals at 20.0 to 40.0 ppm can be observed that are supposed to adsorption. Considering the total Brønsted acid sites (about be oligomerized compounds27. Therefore, the 85.0 ppm signal 1.03 mmol/g) on the catalyst, the amount (less than 0.23 mmol/g) should not be ascribed to the oligomeric alkoxy species. Instead, of DEE adsorbed on Brønsted acid sites should be lower than that we proposed on the basis of its chemical shift the formation of (0.34 mmol/g) of TEO at this reaction condition. At temperature alkyl-substituted oxonium ions. Munson and Haw reported the above 300 °C (Fig. 2d, e), the 85.0 ppm signal is almost formation of trimethyloxonium ion (TMO) with a characteristic unobservable, which can be ascribed to the decomposition of 13C NMR signal at 80.0 ppm by the reaction of dimethyl ether on TEO. In the meantime, the secondary reaction of ethene produces fl H-ZSM-528, and they also confirmed that TMO was not an the dominating C3+ hydrocarbons, re ected by the strong high- intermediate in the methanol to hydrocarbons conversion. Since field signals below 30.0 ppm. The signals at 147.3–156.3, 243.6–255.7 and 47.6 ppm are characteristics of the cyclic cations DEE is readily formed and in equilibrium with ethanol (Fig. 1b), – we assume that a triethyloxoium ion (TEO) may be formed. such as dimethylcyclopentenyl and ethylcyclopentenyl ions30 32, To gain insights into the structure of TEO, a 13C–13C J-based which are usually involved in the formation of aromatics. This refocused INADEQUATE (Incredible Natural Abundance Dou- implies that the aromatics-based cycle prevails in a similar route ble Quantum Transfer Experiment)29 MAS NMR experiment was to the conversion of methanol to hydrocarbons33,34. performed, which provides an unambiguous identification of bond connectivity of carbon species. The INADEQUATE 13 13 spectrum (Fig. 2) of H-ZSM-5 reacted with CH3 CH2OH at Reactivity and intermediate role of TEO. In order to explore the fl 13 13 200 °C for 4 s exhibits a clear correlation peak pair between the reactivity of TEO, continuous- ow CH3 CH2OH conversion methylene carbon at 85.0 ppm and the methyl carbon at over H-ZSM-5 was investigated by in situ 13C NMR spectroscopy 12.0 ppm, confirming the formation of TEO that is composed (see methods section for details of the in situ NMR experiments). of ethyl groups. Two other correlations are identified, showing the Figure 3 shows the real-time monitoring of formed surface spe- connectivity of the methylene carbons of DEE (69.0 ppm) and cies on H-ZSM-5 in the ethanol dehydration process. At a con- ethanol (62.5 ppm) to the corresponding methyl carbons at 12.8 stant reaction temperature of 220 °C (Fig. 3a), the appearance of and 17.0 ppm, respectively. The structure of TEO was also 85.0 ppm signal after 200 s points to the formation of TEO. The theoretically optimized in the H-ZSM-5 channel by periodic broad signal centered at 58.5 ppm should come from the adsor- density functional theory (DFT) calculations (Supplementary bed ethanol, whose methyl group produces a high-field strong Fig. 7). The 13C chemical shift of the methylene carbon of TEO is signal at 15.0 ppm. Note that the methylene group of DEE predicted to be 84.6 ppm, in good agreement with the experi- (69.0 ppm) is invisible in the in situ experiment. This is probably mental value. Therefore, our NMR experiments and theoretical due to the strong adsorption of DEE on zeolite, leading to large calculations provide solid evidence for TEO formation in the anisotropic interactions (chemical shifts and nuclear dipole- ethanol dehydration process. dipole couplings) which cannot be efficiently removed by the low For the adsorbed species formed on H-ZSM-5 after reaction at magic spinning speed (2 kHz). However, the formation of DEE is 200 °C for 4 s, the amount of TEO (85.0 ppm), ethoxy (72.8 ppm), evidenced by the 12.5 ppm signal due to its methyl group featured DEE (69.0 ppm) and ethanol (62.5 ) was determined by 13C NMR by high mobility. TEO is observable with increasing the reaction experiment to be 0.34, 0.46, 28.76 and 6.99 mmol/g, respectively. time up to 1200 s (20 min), indicative of its stability at the moderate reaction temperature. The GC analysis of the effluent

1 35 products obtained from the reaction with time-on-stream shows + 2 O 4 O 6 OH the formation of ethene and a small amount of higher hydro- 13 4 carbons (Supplementary Fig. 8). Figure 3b shows the in situ C 2 NMR spectra obtained at elevating temperatures. The TEO is 3 6 1 5 observable over a wide temperature range. The formation of hydrocarbons prevails at higher temperatures as reflected by a set of high-filed signals (8.7–32.6 ppm) while the TEO is significantly reduced due to its decomposition, consistent with the ex situ NMR result (Fig. 1). In comparison, the structurally similar 70 species TMO formed on H-ZSM-5 was readily decomposed (17.0,79.5) 35,36 (62.5,79.5) before the onset of hydrocarbon synthesis . To confirm the intermediate role of TEO in the formation of 80 ethene, the 13C NMR spectra of trapped species obtained from the 13 pulse-quench reactions of CH3 CH2OH were recorded during (69.0,81.8) (12.8, 81.8) the reaction course of 4~64 s at 250 °C. Figure 4a shows the 90 expanded spectra and their deconvolutions (see Supplementary Fig. 9 for the whole spectra), in which the formation and evolution (85.0,97.0) (12.0,97.0) 100 of TEO (85.0 ppm) is evident. We know from above experiments that the ethoxy species (72.8 ppm) is simultaneously produced, 1 13 Double quantum frequency (ppm) Double which is confirmed by 2D H– C HETCOR MAS NMR (Supplementary Fig. 6). Comparison of the integrated 13C signal 100 80 60 40 20 0 intensities shows that the ethoxy species keeps a similar evolution Single quantum frequency (ppm) trend with TEO (Supplementary Table 1). Thus, the formation of surface ethoxy species is most likely related to TEO. Furthermore, Fig. 2 Identification of triethyloxoium ion by correlation NMR spectroscopy. TEO was deliberately prepared by an ion exchange of triethylox- 2D 13C–13C INADEQUATE MAS NMR spectrum of trapped species onium tetrafluoroborate with H-ZSM-5. The formation of surface 13 13 obtained from the reaction of CH3 CH2OH on H-ZSM-5 at 220 °C for 4 s TEO on H-ZSM-5 is confirmed by its characteristic signal at

NATURE COMMUNICATIONS | (2019) 10:1961 | https://doi.org/10.1038/s41467-019-09956-7 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09956-7 a These results demonstrate a direct correlation of TEO to ethoxy 12.5 species and ethene product. 15.0 85.0 58.5 t(s) t:840s Intensity 4 Mechanism of ethene formation via TEO. Our experimental 1200 ×10 observations indicate a strong potential of TEO serving as an 2 active intermediate in the formation of ethene. A whole reaction 1000 network for ethanol dehydration to ethene was proposed (Fig. 5), which includes the previously theoretically predicted unim- 1.5 800 olecular and bimolecular processes17,19 and our proposed TEO- mediated routes. It consists of three types of mechanisms: direct 600 1 ethanol dehydration to ethene (route 5, 6, 7 and 8), diethyl ether decomposition to ethene (route 3 and 4) and TEO-mediated 400 ethene formation (route 1 and 2). DFT calculations were per- 0.5 formed to explore the most favorable ethene formation route and 200 the optimized transition state structures on the zeolite cluster 0 model (Supplementary Fig. 12) are displayed in Supplementary Fig. 13. Our calculations show that all the kinetic steps of ethanol 0 150 100 50 0 –50 dehydration are endothermic processes (Supplementary Table 2). δ (ppm) The calculated Gibbs free energy barrier (Supplementary Fig. 14– 18 for the energy diagrams), pre-exponential factor, and rate fi b 8.7–32.6 coef cient at 473 K of the forward and reverse reactions are tabulated in Table 1. The comparison of energy barriers and rate 85.0 58.5 T (°C) T :250°C Intensity coefficients indicates that the direct dehydration processes (steps ×105 260 9, 10, 15 and 16) are energetically less favorable for ethene for- 4 mation, in agreement with the previous work17. Compared with the direct ethanol dehydration routes (route 240 5, 6, 7 and 8), the two molecules mediated route 4 including fi Δ = = 3 dimer-mediated etheri cation ( Gact(f) 129.0 kJ/mol, kf 5.54 × 10−2 s−1, step 4) and the following DEE decomposition 220 Δ = −1 = −3 −1 to ethene ( Gact(f) 136.1 kJ mol , kf 9.24 × 10 s ,step 2 11) is favorable for ethanol dehydration to ethene, in agreement 200 with the previous report that the reaction path via DEE is preferentially involved in ethene formation at temperatures 1 lower than 500 K17. It is reasonable to assume that TEO would 180 be readily formed due to the facile formation of DEE. This is confirmed by a free energy barrier of ca. 94.1 kJ mol−1 for the 0 formation of TEO from DEE (step 6), which is much lower than 160 Δ = that of DEE decomposition routes ( Gact(f) 136.1 kJ/mol for 150 100 50 0 –50 Δ = −1 δ (ppm) step 11 in route 4 and Gact(f) 155.9 kJ mol for step 12 in route 3). We have experimentally observed that the ethoxy Fig. 3 Real-time NMR monitoring of ethanol dehydration on zeolite. In situ species and DEE can be produced from TEO on H-ZSM-5 13 13 13 C MAS NMR spectra of CH3 CH2OH conversion on H-ZSM-5 with (Fig. 4). This can be understood by the fact that TEO ethylates time on stream (a) and at elevating temperatures (b). Spectra (a) were the conjugate base site of zeolite to form a framework-bound recorded at every 30 s from 0 to 1200 s at 220 °C and 30 scans were ethoxy species with the release of a DEE, which shows the accumulated for each time slice. Spectra (b) were recorded at every 10 °C characteristic property of trialkyloxonium ions acting as an from 160–260 °C and 900 scans were accumulated for each temperature powerful alkylating agent37. This process is calculated to have slice. The 1D spectra on the top are recorded at 840 s (a) and 250 °C (b), an activation energy of 73.6 kJ mol−1 (step 7, Supplementary respectively Table 2), in good agreement with our experimental value (77.3 kJ mol−1), strongly supporting the formation of ethoxy 85.0 ppm in the 13C NMR spectra (Fig. 4b). The concentration of via TEO intermediate. The low free energy barrier of step 8 − TEO determined by 13C MAS NMR is about 1 mmol/g, (107.6 kJ mol 1) suggests the transformation of ethoxy comparable to the amount (1.03 mmol/g) of Brønsted acid site. species to ethene can readily occur. Moreover, the calculated Importantly, the isolation of stable TEO at room temperature electronic energy barrier for the conversion of ethoxy species to − allows us to unambiguously trace its reactivity and transformation ethene (127.6 kJ mol 1, Supplementary Table 2) is similar to − on the catalyst. After heating, TEO was partially converted into the experimentally estimated value (ca. 122 kJ mol 1)forthe ethoxy species (at 72.8 ppm) and DEE (at 69.0 ppm) at lower decomposition of ethoxy species to ethene on H-ZSM-538.The temperature of 80 °C, while higher temperature (200 °C) led to its direct decomposition of TEO into ethene (step 13 in route 2) is complete conversion. We also obtained the activation energy for also considered, which is however characterized by a higher free − the formation of ethoxy species from TEO by measuring the rate energy barrier (127.7 kJ mol 1). A full comparison of free constants at different temperatures using 13C MAS NMR energy barriers and rate coefficients of the rate-determination spectroscopy (Supplementary Fig. 10). The experimentally steps in the proposed reaction network indicates that route 1 determined activation energy is ca.77.3 kJ/mol. It is worth noting (steps 4, 6, 7, 8) and route 2 (steps 4, 6, 13) involving TEO that a large quantity of ethene was observed in the effluent product intermediate are the most favorable routes. Thus, concerning upon the consumption of TEO by GC (Supplementary Fig. 11). the overall kinetics during the ethanol dehydration to ethene,

4 NATURE COMMUNICATIONS | (2019) 10:1961 | https://doi.org/10.1038/s41467-019-09956-7 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09956-7 ARTICLE

a b 72.8 69.0 62.5 85.0 69.0 72.8 250°C, 32s ×4

80 60 ppm 14.0 250°C, 16s 200 °C

×4 69.0 72.8 80 °C 250°C, 8s ×4 12.6 85.0 RT 250°C, 4s ×4

90 70 50ppm 100 50 0 ppm

Fig. 4 Evolution of TEO in the formation of ethene. 13C CP MAS NMR spectra of trapped species on H-ZSM-5 obtained from pulse-quench reactions 13 o 13 of CH3 CH2OH at 250 C for different time (a) and C MAS NMR spectra of TEO-ZSM-5 at room temperature (RT), heated at 80 °C for 5 min, and at 200 °C for 5 min (b). The deconvoluted components are indicated in color

ZeOH H O O +C (1) 2 H (9) Si AI Si –C O ZeOH A 4 2 H + 2 H –H + EtOH O O 4 2 +H O O H H Si AI Si – EtOH Ethene * O O (2) - + EtOH H Si AI Si O H 4 (16) 2 2 (10) O –C O – EtOH O –H 2 O H 4 O B –EtOH –H 2 Si AI Si 2 2 Ethene * +C O +H H +H 2 O 2 + +EtOH +H H –H O O H + H O O (14) HH –2H2O O O H H O O Si AI Si +2H O - 2 Si AI Si O - O (15) Si AI Si

E (3)

(8)

(13) +DEE

–DEE (11) – EtOH Ethoxy H + EtOH O O + C H H O - O Si AI Si (12) O - O – EtOH Si AI Si

+DEE + EtOH O –DEE 2

(7) +H O 2

–H (4) + O + O

O - O H DEE* Si AI Si +H H O - O O O – EtOH TEO* 2 O + Si AI Si –H H O 2 + EtOH (6) O - O (5) Si AI Si D

Fig. 5 Proposed catalytic cycle for ethanol dehydration to ethene. The structures of transition states (TS) involved in these reaction steps are depicted in Supplementary Fig. 13. * indicates the adsorbed state

NATURE COMMUNICATIONS | (2019) 10:1961 | https://doi.org/10.1038/s41467-019-09956-7 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09956-7

−1 −1 −1 Table 1 Reaction Gibbs free energy barrier (ΔGact, kJ mol ), pre-exponential factor(A,s ), and rate coefficient (k,s ) at 473 K of the forward (f) and reverse reactions (r) for the elementary step associated with the eight ethene production routes

Elementary step ΔGact(f) ΔGact(r) Af kf Ar kr TS4$ : à 11 −2 12 4 Route 1 C DEE H2O (4) 129.0 79.4 6.43 × 10 5.54 × 10 4.20 × 10 1.70 × 10 TS6$ : à 13 2 11 5 D TEO H2O (6) 94.1 65.0 1.19 × 10 3.95 × 10 8.68 × 10 6.45 × 10 TS7 TEO Ã$Ethoxy:DEE Ãð7Þ 81.7 66.6 1.29 × 1012 9.61 × 103 9.33 × 1011 4.38 × 105 TS8 Ethoxy $ Etheneà (8) 107.6 75.3 1.32 × 1013 1.30 × 101 1.91 × 1011 4.74 × 104 TS4$ : à 11 −2 12 4 Route 2 C DEE H2O (4) 129.0 79.4 6.43 × 10 5.54 × 10 4.20 × 10 1.70 × 10 TS6$ : à 13 2 11 5 D TEO H2O (6) 94.1 65.0 1.19 × 10 3.95 × 10 8.68 × 10 6.45 × 10 TS13 TEO Ã$Ethene:DEE Ãð13Þ 127.7 78.4 2.23 × 1013 7.81 × 10−2 9.32 × 109 2.17 × 104 TS4$ : à 11 −2 12 4 Route 3 C DEE H2O (4) 129.0 79.4 6.43 × 10 5.54 × 10 4.20 × 10 1.70 × 10 TS12 DEE Ã$Ethoxy:EtOH Ãð12Þ 155.9 78.2 7.12 × 1012 6.01 × 10−5 5.11 × 1011 2.26 × 104 TS8 Ethoxy $ Etheneà (8) 107.6 75.3 1.32 × 1013 1.30 × 101 1.91 × 1011 4.74 × 104 TS4$ : à 11 −2 12 4 Route 4 C DEE H2O (4) 129.0 79.4 6.43 × 10 5.54 × 10 4.20 × 10 1.70 × 10 TS11 DEE Ã$Ethene:EtOH Ãð11Þ 136.1 58.9 1.03 × 1014 9.24 × 10−3 5.11 × 1011 3.06 × 106 TS9$ : Ãð Þ 13 −4 12 5 Route 5 A Ethoxy H2O 9 146.3 69.8 6.75 × 10 6.81 × 10 4.79 × 10 1.91 × 10 TS8 Ethoxy $ Etheneà (8) 107.6 75.3 1.32 × 1013 1.30 × 101 1.91 × 1011 4.74 × 104 TS10$ : Ãð Þ 13 −6 10 1 Route 6 A Ethene H2O 10 171.8 103.3 4.28 × 10 1.04 × 10 2.96 × 10 3.81 × 10 TS15$ : à 14 −8 11 5 Route 7 E Ethene 2H2O (15) 184.8 64.7 1.15 × 10 3.89 × 10 1.71 × 10 6.98 × 10 TS16$ : : Ãð Þ 13 −4 10 −1 Route 8 B Ethene H2O EtOH 16 145.2 119.2 5.37 × 10 9.01 × 10 1.89 × 10 6.76 × 10

the presence of TEO provides an energetically preferable shown that TEO tends to alkylate basic site (Si–O−–Al) of zeolite process at lower reaction temperature. to form surface ethoxy species, which in nature resembles the alkyloxonium salt featured by strong alkylating ability in organic synthesis. This result indicates the reactivity of alkyloxonium ion Discussion could be altered by the zeolite catalyst, which leads to different The dehydration of ethanol to ethene over H-ZSM-5 has been reaction routes. Although the alkyloxonium ion (i.e., a protonated investigated by combined experiments and DFT calculations. Our alcohol) has long been proposed as a key intermediate in the ex situ and in situ solid-state NMR data provide the evidence for dehydration of alcohol to produce alkenes and ethers particularly the formation of TEO in the dehydration process. Generally, in strong liquid acid systems, the detailed characterizations of onium ions that are often isolable as salts are well recognized as TEO in the ethanol dehydration process in this work provide an reaction intermediates37. It was reported that trimethyloxonium example of uncovering the important role of this kind of inter- can be generated on H-ZSM-5 zeolite. However, mechanistic mediate in the formation of ethene on zeolite catalysts, which significance of trimethyloxonium is uncertain since its inter- may open avenues for further experimental and theoretical mediate role for the initial C–C bond formation in methanol exploration of the oxonium ions chemistry in alcohol conversion. conversion was excluded based on experimental observations27,28. Herein, TEO is formed as a stable surface species on the working Methods catalyst in both continuous-flow and pulse-quench reactions. We X-ray power diffraction (XRD). The ammonium form ZSM-5 (Si/Al = 11.5, calculated the host-guest interaction free energy of TEO confined obtaind from Zeolyst) was calcined in air at 823 K for 5 h to obtain the proton form in the H-ZSM-5 channel, which is as low as −360.2 kJ mol−1 at H-ZSM-5. The structure and crystalline nature of H-ZSM-5 zeolites were examined by X-ray diffractometer (X’Pert3 Powder) using a CuKα radiation with a step of 473 K, indicating that the zeolite framework can stabilize TEO 0.02° at a respective voltage of 40 kV and a current of 40 mA. intermediate effectively. The implications of the observation of TEO in ethanol dehy- FT-IR of pyridine adsorption. The FT-IR of pyridine adsorption measurements dration to ethene in present work are manifold. The ethoxy were performed on a Bruker Tensor 27 spectrometer. The catalysts were first species is often experimentally observed during ethanol dehy- activated at 773 K under high vacuum (<10−5 Pa) for 12 h. After cooling down to dration on zeolite, which is generally considered as the precursor 313 K, a background spectrum was collected. Excessive amount of pyridine was then introduced to the infrared cell and held for 2 h to allow equilibrium. The to ethene in both processes of unimolecular dehydration of residual pyridine was removed by vaccum. The FT-IR spectra of pyridine-adsorbed ethanol and bimolecular dehydration of ethanol followed by DEE samples were measured at 313 K after evacuation at 373, 473, 573, 673 and 773 K, decomposition. DEE is largely formed in the initial stage of respectively. dehydration of ethanol and its coverage on acid sites is estimated to be comparable to TEO at lower temperature (e.g., 473 K). The Temperature-programmed desorption of ammonia (NH3-TPD). The NH3-TPD presence of TEO provides an energetically favorable mechanism measurement was performed using a FINESORB-3010 chemisorption instrument. to link DEE and ethoxy species, which is viable for the formation Typically, 100 mg of sample was pre-treated at 423 K for 1 h under 30 sccm of helium gas. After the sample cooled down to 323 K, NH3 was introduced and held of ethene at low temperature. In the onium-ylide mechanism for 1 h. The temperature was then elevated from 323 to 1023 K at a ramping rate of proposed for the C–C bond formation from C1 compound such 10 K/min and the desorbed NH3 was detected by a thermal conductivity as methanol and DME on zeolite, Olah et al. predicated the detector (TCD). intermediate formation of ethyldimethyloxonium, which could undergo β-elimination to yield ethene and DME39. Analogously, Catalytic testing. The H-ZSM-5 powder was pressed into pellets between 60–80 mesh. The pellets (0.2 g) were activated at 400 °C in flowing helium for 1 h the direct deprotonation and decomposition of TEO could lead to prior to the ethanol dehydration reactions. A pulse-quench reactor was used to ethene and DEE on zeolites (route 2), which, however, occurs quench the reaction by reducing the reaction temperature with liquid nitrogen with a relatively higher energy barrier. We have experimentally within a very short period (<1 s)25. Typically, when the reaction proceeded in a

6 NATURE COMMUNICATIONS | (2019) 10:1961 | https://doi.org/10.1038/s41467-019-09956-7 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09956-7 ARTICLE pulse-quench reactor for a pre-set period, the reaction was thermally quenched by All the in situ 13C solid-state NMR experiments were performed at 11.7 T on a pulsing liquid nitrogen onto the catalyst bed, which was achieved by using high- Bruker Avance III 500 spectrometer, equipped with a 7 mm H/X MASCAT probe speed valves controlled by GC computer. In each case, 10 μl of reactant was pulsed (Fig. S19), with resonance frequencies of 500.58 and 125.87 MHz for 1H and 13C, into the reactor (heated at 160–350 °C) containing 0.2 g H-ZSM-5 and allowed to respectively. The magic angle spinning rate was set to 2 kHz. Prior to the in situ react for different time, before quenching by liquid nitrogen. The trapped surface NMR experiments, 0.2 g of pre-dehydrated H-ZSM-5 was pressed into a 7 mm species were analyzed by 13C solid-state NMR spectroscopy and the effluent pro- NMR rotor. An axial hole of 2.5 mm in diameter was made into the sample with a ducts were on line determined by GC-MS analysis. For the continuous flow special tool, by which an uniform annular catalyst bed was formed in the rotor. reaction, ethanol with a weight hourly space velocity (WHSV) of 2 h−1 was reacted After inserting the injection tube into the MAS rotor, the rotor was heated at over the H-ZSM-5 (0.2 g) pellets in a fixed bed reactor at the temperature range of 280 °C for 1 h via the bearing gas, keeping helium (200 ml min−1) injected into the 140–260 °C. rotor, then cooled down to the reaction temperature. 13C labelled ethanol 13 13 13 ( CH3 CH2OH, 98% C, Sigma-Aldrich) was diluted to 50% (v/v) with ethanol 13 fl fl in natural aboundance. The diluted C labelled ethanol was fed into the NMR Gas chromatography. The ef uent products drawn from the owing gas were rotor by the helium carrier gas (weight hourly space velocity (WHSV) = 2.2 h−1) analyzed quantitatively by online GC-FID chromatograph (Shimadzu GC-2010 through a saturator. The reactants flow inside the rotor via the injection tube and fl TM plus) which equipped with a ame-ionization detector and a Supelco Supel-Q pass through the catalyst from the bottom to the top and leave the rotor via an PLOT capillary column (30 m × 0.32 mm × 15 μm). The initial temperature pro- 13 −1 exhaust tube in the rotor cap. The C CP/MAS NMR spectra were recorded with a gramming started at 40 °C (maintained for 2 min), followed by a rate of 5 °C min contant time of 5 ms and a recycle delay of 1 s. 30 and 900 scans were accumulated −1 fi to 132.5 °C and a rate of 10 °C min to the nal temperature of 220 °C. The for each spectrum at different time and temperature respectively. The 13C chemical retained products in the catalyst was directly analyzed by solid-state NMR spec- shifts were referenced to HMB (a second reference to TMS). The in situ 13C MAS troscopy (see the following). NMR spectra in Fig. 3 were generated by the overlay of 1D NMR spectra as a function of reaction time or temperature. The correlation between 13C resonances Preparation of triethyloxoium ion (TEO) on H-ZSM-5. A concentration of 0.2 g and the reaction parameters (time and temperature) was followed in the 2D maps, of H-ZSM-5 zeolite (Si/Al = 11.5) was dehydrated on a vacuum line (<10−3 Pa) at which facilitates the monitoring of the fate of the organic species during the time or 400 °C for 12 h. After dehydration, the sample was allowed to cool down to room temperature evolutions. temperature for subsequent use. A concentration of 0.3 g triethyloxonium tetra- The activation energy of the formation of ethoxy species from TEO on H-ZSM- fluoroborate (Aldrich) was added into 3 g dry dichloromethane solvent, and then 5 was measured as follows. The sealed NMR rotor containing the TEO-ZSM- fi the dehydrated zeolite sample was added into the solution. All the operations were 5 sample was heated at a speci c temperature (from 328 to 348 K) for a period of 13 fi time, and then the reaction was quenched by liquid N2. C MAS NMR carried out in a glovebox lled with pure N2. After ultrasonic treatment of the mixture in ice water bath at 0 °C for 20 min, the mixtrue was filtrated and evac- measurement was performed at room temperature. At such low reaction uated completely and the sample was dried at room tempearture for 5 h by temperature, only ethoxy species and diethyl ether were produced on the TEO- vacuum. The obtianed triethyloxoium ion (TEO) exchanged ZSM-5 zeolite was ZSM-5. The concentration of TEO during the reaction was measured by denoted as TEO-ZSM-5. normalized integrated NMR signal at 85 ppm. Finally, the temperature dependent rate constant was obtained and the activation energy was derived from the Arrhenius equation. Solid-State NMR experiments. After the reaction was quenched, the pulse- quench reactor containing the catalyst was sealed. The sealed reactor was then fi transferred to a glove box lled with pure N2 and the catalyst was packed into to an Theoretical calculations. The host-guest interactions between the cations and NMR rotor for ex situ NMR measurements. To enhance the detection sensitivity, zeolite framework would result in the 13C chemical shift moving due to the re- 13 13 C labeled ethanol was used for C NMR experiments in both the pulse-quench distributions of electronic environment invoked by the H-ZSM-5 confined pore. and the in situ reactions. Thus, to unambiguously assign the experimental NMR results, the 13C NMR All the ex-situ 13C solid-state NMR spectroscopy experiments were carried out chemical shifts of the cations are further calculated using the periodic boundary at 9.4 T on a Bruker Avance III-400 spectrometer, equipped with a 4 mm probe, condition. with resonance frequencies of 399.33 and 100.42 MHz for 1H and 13C, respectively. The geometry optimizations were performed by using the CASTEP program41 The magic angle spinning rate was set to 10 kHz. For the 1H → 13C CP/MAS NMR with the generalized gradient approximation (GGA) proposed by experiments, the Hartmann-Hahn condition was achieved using Perdew–Burke–Ernzerhof (PBE) functional. And the ultrasoft pseudo potential, hexamethylbenzene (HMB), with a contact time of 5 ms and a repetition time of fine plane wave cut-off energy (340 eV) and a default fine level Monkhorst-Pack K- 2 s. The hpdec 13C MAS NMR experiments were performed using a 13C 90-degree point (1 × 1 × 1) were adopted to sample the Brillouin zone. DFT-D method42 was pulse length of 5 μs and a recycle delay of 5 s. The 13C chemical shifts were used in the structure optimization and the NMR calculation to accurately describe referenced to HMB (a second reference to TMS). the weak interaction in H-ZSM-5. During optimization, the 22T active site atoms The 27Al MAS NMR spectra were also acquired on the same 4 mm probe by and the adsorbed cation were relaxing to their equilibrium positions. All of the small-flip angle technique with a pulse length of 0.3 μs(<π/12) and a recycle delay NMR shielding calculations were performed by the GIPAW method in the MS of 1 s. The magic angle spinning rate was set to 10 kHz. The 27Al chemical shifts CASTEP-NMR code at the GGA/PBE level based on the optimized zeolite 29 were referenced to 1 M Al(NO3)3 aqueous solution (0 ppm). Si MAS NMR structures. All of the 13C chemical shifts (δ 13C) cal were derived using the experiments were carried out at 7.1T on a Varian Infinity plus-300 spectrometer, CASTEP-NMR module available in the Materials Studio package based on the with resonance frequencies of 299.78 and 70.11 MHz for 1H and 29Si, respectively. optimized structures of cation accommodation in the H-ZSM-5 zeolite. The fine K- Single-pulse 29Si MAS NMR spectra with high power proton decoupling were point (1 × 1 × 1) and cut-off energy of 550 eV were employed, combined with recorded on a 7.5 mm probe, using a π/2 pulse of 5 μs, a recycle delay of 80 s and a core–valance interactions described by ultrasoft pseudo potential generated on-the- spinning rate of 4 kHz. The 29Si chemical shifts were referenced to kaolinite fly. The 13C calculated chemical shifts were further converted to (δ 13C) cal values, (−91.5 ppm). which were referred to the absolute shielding of TMO, namely 80 ppm for the 2D 1H –13C CP HETCOR experiment was performed using a Avance III experimental values. 800 spectrometer operating at a 1H Larmor frequency of 800.36 MHz and a 4 mm For the ethene formation pathway calculation, H-ZSM-5 zeolites are HCN E-free probe at a spinning frequency of 12 kHz. The 1H π/2-pulse length was represented by a 72T model (HSi71AlO179, 252 atoms), which were extracted from 4.85 μs. 13C magnetization was created using a cross-polarization (CP) ramp of their crystallographic structural data (http://www.iza-structure.org/databases/). The magnitude 80 to 100% with a contacted time of 5 ms. 256 transients were co-added 72T contains the complete double 10-MR intersection pores of H-ZSM-5 zeolite. using a recycle delay of 2 s. A total of 256 t1 FIDs were recorded at increments of The terminal Si–H was fixed at a bond length of 1.47 Å, oriented along the 3.19 ms using the States-TPPI method to achieve sign discrimination in F1. 1H direction of the corresponding Si–O bond. The substituted Al atom was placed at 40 fi SPINAL-64 decoupling was applied during the t2 acquision with a RF- eld the T12 site of the crystallographic position during structural optimization, amplitude of 51.5 kHz. whereas the proton was located at the O24 site. The previous works have 13 13 29 2D C- CCPJ-refocused INADEQUATE MAS NMR experiment was demonstrated that the Si12-O24(H)-Al12 Bronsted acid site located at the channel carried out using a Avance III 800 spectrometer at a MAS speed of 8 kHz. The 1H intersection of H-ZSM-5 zeolite with maximum accessibility for bulky reactants π/2-pulse length was 2.75 us. 13C magnetization was created using a cross- and transition states17,43,44. The 22T active site atoms (HSi21AlO25, 48 atoms) and polarization (CP) ramp of magnitude 80 to 100% with a contacted time of 7 ms. the adsorbed hydrocarbon complex were treated as the high-layer (See Fig. S12) 13C π/2 and π pulses were 4.8 us and 9.6 µs, respectively. The J-evolution τ periods while the rest of the frameworks were treated as the low-layer. To retain the were rotor synchronized and set to 3.24 ms. 64 transients were co-added using a structural integrities of the modeled zeolite, partial structure optimizations of the recycle delay of 3 s. A total of 160 t1 FIDs were recorded at increments of 3.97 ms 72T cluster were performed by relaxing the atoms in the the high-level layer while using the States-TPPI method to achieve sign discrimination in F1. 1H SPINAL-64 keeping the rest of atoms fixed at their crystallographic positions. All the TS decoupling with a RF-field amplitude of 90.91 kHz was employed after CP covering structures are found by the QST 3 method in the Gaussian program. Then the IRC all t1, J-evolution and t2 periods. The two frequency axes in the 2D spectra are used (Intrinsic Reaction Coordinate) method was used to determine the structures of the to assign the 13C resonances corresponding to through-bond 13C-13C corresponding reactant and product. Then based on the imaginary vibrational connectivities, which allows for unambiguous assignment and structure model of the optimized TS, we adjusted the positions of the vibrational atoms determination of organic compounds. slightly along the calculated reaction coordinate on the two directions toward the

NATURE COMMUNICATIONS | (2019) 10:1961 | https://doi.org/10.1038/s41467-019-09956-7 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09956-7 reactant and product, respectively, and finally optimized the resulting structures to methanol-to-olefins catalysis by theory and experiment. ChemPhysChem 14, the minimum structures. These methods have been widely employed in the 1526–1545 (2013). previous theoretical work. 12. Olsbye, U. et al. Conversion of methanol to hydrocarbons: how zeolite cavity A combined theoretical approach, namely ONIOM (ωB97xd/6–31 G(d,p): am1) and pore size controls product selectivity. Angew. Chem. Int. Ed. 51, was used for the geometry optimization of adsorption states and transition states 5810–5831 (2012). (TS). The ωB97XD hybrid density function was developed to consider long-range- 13. Chiang, H. & Bhan, A. Catalytic consequences of hydroxyl group location on corrected hybrid functional, which implicitly accounted for empirical dispersion the rate and mechanism of parallel dehydration reactions of ethanol over and could describe long-range dispersion interactions well with respect to the acidic zeolites. J. Catal. 271, 251–261 (2010). 42 traditional density functional theory methods . This functional was also recently 14. Phung, T. K. & Busca, G. Diethyl ether cracking and ethanol dehydration: found to perform very well for the description of adsorption and reactions on Acid catalysis and reaction paths. Chem. Eng. J. 272,92–101 (2015). ω – zeolites. All energies report herein were predicted at the B97XD/6 31 G(d, p) 15. Gayubo, A. G., Tarrío, A. M., Aguayo, A. T., Olazar, M. & Bilbao, J. Kinetic level based on the optimized structures. modelling of the transformation of aqueous ethanol into hydrocarbons on a The harmonic frequency calculations employing a partial Hessian vibrational – 45 HZSM-5 Zeolite. Ind. Eng. Chem. Res. 40, 3467 3474 (2001). analysis (PHVA) , including the 22T high layer active acidic sites and organic 16. Kondo, J. N. et al. Activation energies for the reaction of ethoxy species to species were performed to check whether the stationary points found exhibit the ethene over zeolites. J. Phys. Chem. C. 114, 20107–20113 (2010). proper number of imaginary frequencies. In frequency calculations, besides the 17. Alexopoulos, K. et al. DFT-based microkinetic modeling of ethanol atoms in high level and organic fragments, the constraints of the zeolite framework dehydration in H-ZSM-5. J. Catal. 339, 173–185 (2016). were also kept as the same in geometry optimizations, so one negative frequency 18. Kim, S., Robichaud, D. J., Beckham, G. T., Paton, R. S. & Nimlos, M. R. would be observed for transition state point and none for the corresponding Ethanol dehydration in HZSM-5 studied by density functional theory: reactant and product. The Gibbs free energies (G) at 423 K (T) were calculated 119 –3614 (2015). from harmonic frequencies. evidence for a concerted process. J. Phys. Chem. A , 3604 19. Xin, H. et al. Catalytic dehydration of ethanol over post-treated ZSM-5 ¼ À ´ ¼ ðÞÀþ þ þ þ ´ : ð Þ – G H T S E ZPVE Hvib Htrans Hrot T S 1 zeolites. J. Catal. 312, 204 215 (2014). It can be seen that Gibbs free energies (G) include contributions from electronic 20. Wang, W., Jiao, J., Jiang, Y. J., Ray, S. S. & Hunger, M. Formation and energies (E), zero-point vibrational energies (ZPVE), vibrational enthalpies (H ), decomposition of surface ethoxy species on acidic zeolite Y. Chemphyschem 6, vib – translational (Htrans) and rotational enthalpies (Hrot) for reactant molecules and the 1467 1469 (2005). effect of entropy energy (T × S), which can more accurately describe the 21. Kondo, J. N., Ito, K., Yoda, E., Wakabayashi, F. & Domen, K. An ethoxy mechanism of ethanol dehydration reaction. intermediate in ethanol dehydration on bronsted acid sites in zeolite. J. Phys. And then, the reaction rate constants (k) at 473 K were further obtained by Chem. B 109, 10969–10972 (2005). transition state theory (TST): 22. Zecchina, A. et al. IR spectroscopy of neutral and ionic hydrogen-bonded    Δ Δ Δ Δ complexes formed upon interaction of CH3OH, C2H5OH, (CH3)2O, (C2H5)2O Hact kBT Sact Hact kBT Gact k ¼ A: exp À ¼ : exp : exp À ¼ : exp À ; and C4H8O with H-Y, H-ZSM-5 and H-mordenite: comparison with RT h R RT h RT analogous adducts formed on the H-Nafion superacidic membrane. J. Chem. ð2Þ Soc., Faraday Trans. 92, 4863–4875 (1996). ’ ’ 23. Lee, C. C., Gorte, R. J. & Farneth, W. E. Calorimetric study of alcohol and where A is pre-exponential factor, kB is Boltzmann s constant, h is Planck s – constant and T is the reaction temperature. All the geometry optimizations and nitrile adsorption complexes in H-ZSM-5. J. Phys. Chem. B 101, 3811 3817 frequency calculations were performed using the Gaussian 09 package46. (1997). 24. Nguyen, T. M. & Le Van Mao, R. Conversion of ethanol in aqueous solution over ZSM-5 zeolites: Study of the reaction network. Appl. Catal. 58, 119–129 Data availability (1990). All the data supporting the findings of this study are available within the article and 25. Goguen, P. W. et al. Pulse-quench catalytic reactor studies reveal a carbon- its supplementary information file or from the corresponding author upon reasonable pool mechanism in methanol-to-gasoline chemistry on zeolite HZSM-5. J. request. Am. Chem. Soc. 120, 2650–2651 (1998). 26. Stepanov, A. G., Luzgin, M. V., Romannikov, V. N., Sidelnikov, V. N. & Received: 10 November 2018 Accepted: 2 April 2019 Paukshtis, E. A. The Nature, structure, and composition of adsorbed hydrocarbon products of ambient temperature oligomerization of ethylene on acidic zeolite H-ZSM-5. J. Catal. 178, 466–477 (1998). 27. Munson, E. J., Kheir, A. A., Lazo, N. D. & Haw, J. F. In situ solid-state NMR study of methanol-to-gasoline chemistry in zeolite HZSM-5. J. Phys. Chem. 96, 7740–7746 (1992). References 28. Munson, E. J. & Haw, J. F. NMR observation of trimethyloxonium formation 1. Alvira, P., Tomás-Pejó, E., Ballesteros, M. & Negro, M. J. Pretreatment from dimethyl ether on zeolite HZSM-5. J. Am. Chem. Soc. 113, 6303–6305 technologies for an efficient bioethanol production process based on (1991). enzymatic hydrolysis: a review. Bioresour. Technol. 101, 4851–4861 (2010). 29. Lesage, A., Bardet, M. & Emsley, L. Through-bond carbon-carbon 2. Sun, J. & Wang, Y. Recent advances in catalytic conversion of ethanol to connectivities in disordered solids by NMR. J. Am. Chem. Soc. 121, chemicals. ACS Catal. 4, 1078–1090 (2014). 10987–10993 (1999). 3. Zhang, M. & Yu, Y. Dehydration of ethanol to ethylene. Ind. Eng. Chem. Res. 30. Wang, C. et al. New insight into the hydrocarbon-pool chemistry of the 52, 9505–9514 (2013). methanol-to-olefins conversion over zeolite H-ZSM-5 from GC-MS, solid- 4. Diaz Alvarado, F. A. & Gracia, F. Steam reforming of ethanol for hydrogen state nmr spectroscopy, and DFT calculations. Chem. Eur. J. 20, 12432–12443 production: Thermodynamic analysis including different carbon deposits (2014). representation. Chem. Eng. J. 165, 649–657 (2010). 31. Wang, C. et al. Experimental evidence on the formation of ethene through 5. Bi, J. D., Guo, X. W., Liu, M. & Wang, X. S. High effective dehydration of bio- carbocations in methanol conversion over H-ZSM-5 zeolite. Chem. Eur. J. 21, ethanol into ethylene over nanoscale HZSM-5 zeolite catalysts. Catal. Today 12061–12068 (2015). 149, 143–147 (2010). 32. Wang, C. et al. Direct detection of supramolecular reaction centers in the 6. Roy, S. et al. Mechanistic study of alcohol dehydration on gamma-Al2O3. ACS methanol-to-olefins conversion over zeolite H-ZSM-5 by 13C-27Al solid-state Catal. 2, 1846–1853 (2012). NMR spectroscopy. Angew. Chem. Int. Ed. 55, 2507–2511 (2016). 7. Janik, M. J., Macht, J., Iglesia, E. & Neurock, M. Correlating acid properties 33. Svelle, S. et al. Conversion of methanol into hydrocarbons over zeolite H- and catalytic function: a first-principles analysis of alcohol dehydration ZSM-5: Ethene formation is mechanistically separated from the formation of pathways on polyoxometalates. J. Phys. Chem. C. 113, 1872–1885 (2009). higher alkenes. J. Am. Chem. Soc. 128, 14770–14771 (2006). 8. Sun, J. et al. Direct conversion of bio-ethanol to isobutene on nanosized 34. Sun, X. et al. On reaction pathways in the conversion of methanol to ZnxZryOz mixed oxides with balanced acid–base sites. J. Am. Chem. Soc. 133, hydrocarbons on HZSM-5. J. Catal. 317, 185–197 (2014). 11096–11099 (2011). 35. Munson, E. J., Kheir, A. A., Lazo, N. D. & Haw, J. F. Insitu solid-state NMR- 9. Phillips, C. B. & Datta, R. Production of ethylene from hydrous ethanol on study of methanol-to-gasoline chemistry in zeolite HZSM-5. J. Phys. Chem. H-ZSM-5 under mild conditions. Ind. Eng. Chem. Res. 36, 4466–4475 (1997). 96, 7740–7746 (1992). 10. Van der Borght, K. et al. Insights into the reaction mechanism of ethanol 36. Munson, E. J., Kheir, A. A. & Haw, J. F. An in-situ solid-state NMR-study of conversion into hydrocarbons on H-ZSM-5. Angew. Chem. Int. Ed. 55, the formation and reactivity of trialkylonium ions in zeolites. J. Phys. Chem. 12817–12821 (2016). 97, 7321–7327 (1993). 11. Hemelsoet, K., Van der Mynsbrugge, J., De Wispelaere, K., Waroquier, M. & 37. Olah, G. A., Laali, K. K., Wang, Q. & Prakash, G. S. Onium Ions (John Wiley & Van Speybroeck, V. Unraveling the reaction mechanisms governing Sons, New York, 1998).

8 NATURE COMMUNICATIONS | (2019) 10:1961 | https://doi.org/10.1038/s41467-019-09956-7 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09956-7 ARTICLE

38. Kondo, J. N., Yamazaki, H., Osuga, R., Yokoi, T. & Tatsumi, T. Mechanism of performed the DFT calculations. X. Z, J.X, and F.D collected and analyzed the in situ decomposition of surface ethoxy species to ethene and acidic OH groups on NMR spectra; X.Z, C.W., Y.Y.C, J.X and F.D. wrote the manuscript, and all authors H-ZSM-5. J. Phys. Chem. Lett. 6, 2243–2246 (2015). discussed the experiments and final manuscript. 39. Olah, G. A. et al. Onium Ylide chemistry. 1. Bifunctional acid-base-catalyzed conversion of heterosubstituted methanes into ethylene and derived Additional information hydrocarbons. The onium ylide mechanism of the C1.fwdarw. C2 conversion. – Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- J. Am. Chem. Soc. 106, 2143 2149 (1984). 019-09956-7. 40. Fung, B. M., Khitrin, A. K. & Ermolaev, K. An Improved broadband decoupling sequence for liquid crystals and solids. J. Magn. Reson. 142, Competing interests: The authors declare no competing interests. 97–101 (2000). 41. Clark Stewart, J. et al. First principles methods using CASTEP. Z. Krist. Cryst. – Reprints and permission information is available online at http://npg.nature.com/ Mater. 220, 567 570 (2005). reprintsandpermissions/ 42. Chai, J. D. & Head-Gordon, M. Long-range corrected hybrid density functionals with damped atom-atom dispersion corrections. Phys. Chem. – Journal peer review information: Nature Communications thanks the anonymous Chem. Phys. 10, 6615 6620 (2008). reviewers for their contribution to the peer review of this work. 43. Brändle, M. & Sauer, J. Acidity differences between inorganic solids induced by their framework structure. a combined quantum mechanics/molecular ’ – Publisher s note: Springer Nature remains neutral with regard to jurisdictional claims in mechanics ab initio study on zeolites. J. Am. Chem. Soc. 120, 1556 1570 (1998). published maps and institutional affiliations. 44. Svelle, S., Tuma, C., Rozanska, X., Kerber, T. & Sauer, J. Quantum chemical modeling of zeolite-catalyzed methylation reactions: toward chemical accuracy for barriers. J. Am. Chem. Soc. 131, 816–825 (2009). 45. Ghysels, A., Van Neck, D., Van Speybroeck, V., Verstraelen, T. & Waroquier, Open Access This article is licensed under a Creative Commons M. Vibrational modes in partially optimized molecular systems. J. Chem. Phys. Attribution 4.0 International License, which permits use, sharing, 126, 224102 (2007). adaptation, distribution and reproduction in any medium or format, as long as you give 46. Frisch, M. J. et al. Gaussian 09. Gaussian Inc., Wallingford, CT, 2010. appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless Acknowledgements indicated otherwise in a credit line to the material. If material is not included in the This work was supported by the National Natural Science Foundation of China (Grants article’s Creative Commons license and your intended use is not permitted by statutory 21622311, 21473245 and 21603265, 21733013, 21773296), key program for frontier regulation or exceeds the permitted use, you will need to obtain permission directly from science of the Chinese Academy of Sciences (QYZDB-SSW-SLH027) and Hubei Pro- the copyright holder. To view a copy of this license, visit http://creativecommons.org/ vincial Natural Science Foundation (2017CFA032). licenses/by/4.0/.

Author contributions © The Author(s) 2019 X.Z, C.W, and X.L. Z prepared the samples and carried XRD and catalytic testing. Q.W, G.D.Q, and N.D.F performed and analyzed the ex-situ solid state NMR spectra. Y.Y.C

NATURE COMMUNICATIONS | (2019) 10:1961 | https://doi.org/10.1038/s41467-019-09956-7 | www.nature.com/naturecommunications 9